Executive Summary

Contributors

Alex Lesicko, PhD is a postdoctoral fellow studying auditory coding at the University of Pennsylvania. She recently moved to Philadelphia from Chicago, where she completed her PhD in neuroscience.

Jake Riley is a clinical data analyst at the Children’s Hospital of Philadelphia (CHOP). He enjoys developing tools for analytic teams and specializes on data visualization and geospatial information systems (GIS).

Javier Jasso is a certified speech–language pathologist and a PhD candidate in communication sciences and disorders at the University of Texas at Austin. Javier has expertise in the assessment of culturally/linguistically diverse children, focusing on bilingual language acquisition.

Katerina Placek is a PhD candidate in neuroscience at the University of Pennsylvania and a co-organizer of R-Ladies Philly. She enjoys integrating outreach with teaching and learning in the local data science community.

Problem definition and dataset

In this series of analyses, we investigated factors relating to an animal’s trajectory in the PAWS system using PetPoint data from 2018. We operationalized animal trajectory as wait time and outcome (e.g. adoption), with wait time defined as time in days from intake to outcome. We restricted our analyses to dogs and cats, as other animals’ data points were sparse and compromised statistical power. Our primary factors of interest included animal characteristics (size, breed, health), intake type, and seasonal patterns. Through this, we aimed to elucidate factors contributing to an animal’s time in the PAWS system.

Results

Our analyses focused on four facets of the PetPoint dataset:
1. Data Exploration and defining ‘Wait Time’
2. Animal Characteristics
3. Intake and Outcome Characteristics
4. Seasonal/Locational Patterns

1. Data Exploration and Defining ‘Wait Time’

We first examined animals’ outcomes for all 2831 animals in the 2018 Pet Point datset.

Outcome Type Number of Animals
Adoption 2777
Return to Owner/Guardian 30
Transfer Out 24

Next, we calculated ‘wait time’ (time in days from intake to outcome) for each animal:

We found that the median wait time for an animal at PAWS in 2018 was 45 days.

We then analyzed outcome type and wait time based on animal species. We calculated the numbers of animals per species in the PetPoint dataset:

Species Number of Animals
Cat 2424
Dog 407
Wild Mammal 0

And we visualized the frequency of outcome types per species:

Species Outcome Type Median Wait Time (Days) Number of Animals
Cat Adoption 52.0 2390
Cat Return to Owner/Guardian 4.5 18
Cat Transfer Out 18.5 16
Dog Adoption 20.0 387
Dog Return to Owner/Guardian 2.0 12
Dog Transfer Out 1.5 8

And also visualized the frequency of wait time lengths per species:

Species Median Wait Time (Days)
Cat 51
Dog 18

We found that the median wait time for a PAWS cat in 2018 was 51 days and that the median wait time for a PAWS dog in 2018 was 18 days.

Next, for each species, we visualized which PetPoint variables contributed to differences in wait time:

We focused our subsequent analyses on the variables with the greatest contributions to wait times (i.e. the longest horizontal lines on each plot).

2. Animal Characteristics

For animal characteristics contributing to wait time at PAWS, we first examined breed per species:

Most, if not all, cats from PAWS were ‘domestic short hair’, while dogs’ breed variability was greater.

We then classified each animal as having a ‘long’ or a ‘short’ wait time based on the median wait time per species, and examined whether wait time differed on breed for dogs and cats:

Our findings indicate that for dogs, Shih Tzus tend to have shorter wait times whereas Terriers tend to have longer wait times.

Based on the large number of unique dog breeds in the PetPoint dataset, we classified dogs into 3 size categories based on average weight per breed. We then visualized wait times by size category:

Our statistical analyses demonstrated no significant differences in wait time based on dog size category.

We then examined age group per species, to determine whether wait time different on age group for dogs and cats:

We found that cats in younger age groups have longer wait times, whereas but found no differences in dogs based on age group.

Last, we examined health at intake and outcome per species relative to wait times.

Species Intake Asilomar Median Wait Time (Days) Number of Animals
Cat Healthy 47 694
Cat Treatable-Manageable 29 104
Cat Treatable-Rehabilitatable 57 652
Cat Unhealthy/Untreatable 56 7
Dog Healthy 20 73
Dog Treatable-Manageable 26 11
Dog Treatable-Rehabilitatable 20 46

Species Outcome Asilomar Median Wait Time (Days) Number of Animals
Cat Healthy 47 694
Cat Treatable-Manageable 29 104
Cat Treatable-Rehabilitatable 57 652
Cat Unhealthy/Untreatable 56 7
Dog Healthy 20 73
Dog Treatable-Manageable 26 11
Dog Treatable-Rehabilitatable 20 46

Our analyses revealed that for cats only, health condition at intake and outcome was associated with longer median wait time. Specifically, cats classified as ‘Treatable-Rehabilitatable’ and ‘Unhealthy/Untreatable’ had longer wait times than cats classified as ‘Healthy’ or ‘Treatable-Manageable’.

3. Intake Characteristics

Next, we examined intake characteristics across the petpoint dataset.

We first examined the frequency of primary intake type for all animals and calculated the median wait days for all animals based on intake type:

Intake Type Number of Animals Median Wait Time (Days)
Owner/Guardian Surrender 120 15
Return 213 13
Stray 939 46
Transfer In 1513 52

Using analysis of variance, we examined whether wait days differed based on animal intake type:

Our results indicated that animals at PAWS with an intake of “Stray” or “Transfer In” had significantly longer wait times relative to animals who with an intake type of “Owner/Guardian Surrender” or “Return.”

We next examined intake type by species:

Intake Type Number of Cats Median Wait Time (Days)
Owner/Guardian Surrender 33 41
Return 134 15
Stray 771 50
Transfer In 1068 58
Intake Type Number of Dogs Median Wait Time (Days)
Owner/Guardian Surrender 64 12.0
Return 42 10.5
Stray 18 27.5
Transfer In 197 25.0

And we examined whether the effect of intake type on wait time differed by animal species:

We found that as previously, cats had longer wait times for each intake type relative to dogs.

We next examined the intake subtype for all animals:

Intake Subtype Number of Animals Median Wait Time (Days)
ACCT Diversion 55 15.0
Abandoned at Shelter 43 30.0
Driver Pickup 2 39.0
In Person Surrender to Shelter 70 17.5
Non-Partner Trans In 12 62.5
Partner Transfer In 1501 52.0
Police Pickup / Drop Off 1 57.0
Project Meow 452 56.5
Public Drop Off 433 40.0
Returned Adoption 213 13.0
Unidentified Animal in Shelter 3 58.0

We focused our analysis of intake subtype on animals in the Public Dropoff and Returned Adoption categories:

We found that animals with an intake subtype of ‘Public Drop Off’ had a significantly longer wait time than animals with an intake subtype of ‘Returned Adoption.’

4. Seasonal and Locational Patterns

Last, we examined animals’ wait time at PAWS by season and location. We examined the frequency of intakes per month:

And examined the frequency of releases per month:

And calculated the difference between intakes and releases per month:

We found that the summer and spring months have higher intakes relative to releases.

We then looked at monthly trends in intakes and releases by species:

We found that cats have higher intakes than releases in spring and summer months, but observed no seasonal patterns for dogs.

We then examined seasonal patterns in intake-release index by location for each species:

We then examined seasonal patterns in wait time:

And examined seasonal patterns in wait time per species:

We then visualized the wait time by location for each species.

Our findings demonstrated that PAWS offsite adoptions has the shortest wait time for cats.

Our findings demonstrated that the PAC location has the shortest wait time for dogs.

Last, we examined wait time by intake health condition for each species.

For cats, we found a that animals classified as ‘Sick’, ‘Injured’, and ‘Under 7 Weeks’ tended to have longer wait times in the spring and summer months.

For dogs, we found a that animals classified as ‘Injured’ and ‘Sick’ also tended to have longer wait times in the spring and summer months.

Conclusions and Next Steps

In conclusion, our analyses revealed that animal characteristics, intake characteristics, and seasonal and locational patterns contribute to animals’ wait times at PAWS in the 2018 year. Almost all animals’ outcomes from 2018 were adoptions, meaning that PAWS is fulfilling its goal of finding homes for needy animals in the Philadelphia area. The median wait time for cats (51 days) was longer than the median wait time for dogs (18 days), and this was likely due to 1) greater number of cats vs dogs, and 2) longer wait times for sick, young cats in the spring and summer months. We also observed differences in wait time by PAWS location for each species, but this is likely due to the number and species of animals at each location. Overall, our findings indicate that PAWS may want to focus resources on young and sick cats in the spring and summer months in order to reduce wait times.